IMPROVING BEST ESTIMATE APPROACHES WITH UNCERTAINTY QUANTIFICATION IN NUCLEAR REACTOR THERMAL-HYDRAULICS

The purpose of the present doctoral research is to improve on two issues identified in the frame of use of Best Estimate Plus Uncertainty (BEPU) approach in nuclear safety analysis: validation and assessment of the computational codes and evaluation of uncertainty of code input parameters. The application of best estimate (BE) methodology to nuclear reactor technology and, in particular to the safety analysis within the licensing process, implies availability of mature and qualified computer codes (e.g. SYS-TH, CFD, etc.) that are able to simulate accurately a wide spectrum of complex single- and two-phase flows and heat transfer phenomena envisaged to occur in Light Water Reactor (LWR) systems under normal, off normal and accidental conditions. Therefore, a rigorous assessment of computational codes against various experiments is needed in order to demonstrate their qualification level and, thus to ensure the validity of performed BEPU analysis within an assigned process. The contribution provided by this work consists in extended validation of SYS-TH codes against experimental data available from numerous separate effect test facilities (SETF) representing different accident scenario in Pressurized Water Reactor (PWR) systems (e.g. large break loss of coolant accident, main steam line break, etc.), related phenomena and physical mechanisms (reflood, in-vessel coolant mixing under asymmetric buoyant flow conditions, etc.). In this context, a number of reflood tests from FEBA, ACHILLES and THETIS experiments were selected and used in order to challenge a 1-D and 3-D models of best estimate system thermal-hydraulic code CATHARE with different operating conditions (e.g. pressure, inlet temperature, liquid velocity, etc.) and geometrical configurations of the fuel assemblies (with and without ballooned rods). Rigorous sensitivity studies to the most influential parameters, accuracy evaluation and uncertainty quantification of the code predictions have been performed. On the other side, the capabilities of 3-D CATHARE model to predict in-vessel flow (IVF) mixing under asymmetric buoyant coolant flow conditions have been assessed using experimental data from ROCOM experiments. The main goal was to respond to the growing interest of nuclear industry to model NPP core using three-dimensional features of thermal-hydraulic system codes by performing a thorough check upon the validity and accuracy of such computational tools. In particular, the relevant modelling issues were identified and discussed, so as to point out the main capabilities and limitations in the present state-of-the-art tools and methods. Additionally, the application of CFD codes to support the set-up of best estimate 3-D SYS-TH nodalization of reactor pressure vessel was successfully demonstrated. This was achieved by using the ANSYS CFX code to evaluate the pressure losses throughout the vessel, and with subsequent application of additional loss coefficients in the CATHARE reference model so as to match the pressure drops predicted by the CFD model. Furthermore, the research contributes to further extending the basis for the use of CMFD codes for nuclear reactor applications by performing simulations of two-phase flows in different geometrical configurations. There is a variety of possible applications of CMFD codes to the NPP-relevant safety issues, such as two-phase pressurized thermal shock, critical heat flux, pool heat exchanger, spray systems in containment, and/or advanced design concepts, such as passive safety options, design optimization, etc. However, additional efforts are still needed in order to fulfil the quality assurance requirements that will make such tools applicable to the nuclear reactor technology and, in particular to the safety analysis within the licensing process. Therefore, to address this issue an extensive validation activity of the NEPTUNE_CFD 2.0.1 code against experimental data on void fraction distribution in vertical channel and subchannel configurations has been performed (based mainly on PSBT tests). The application of BE computer codes and models implies the evaluation of uncertainties. This is connected with the imperfect nature of the codes (e.g. model deficiencies, approximations in the numerical solution, etc.) and of the code application process (e.g. use of specific models, nodalization effects, etc.). Outcomes from the international benchmarks (like BEMUSE and PREMIUM) showed that the use of engineering judgment in the process of selection of influential code input parameters (IP) and the imperfect knowledge of the code IP uncertainties considerably affects the results of performed uncertainty analysis. Therefore, the last part of this doctoral research aims at contributing to the issue of uncertainty quantification of system thermal-hydraulic code input parameters. A procedure for seeking of the variation ranges of selected IP based on multi-parameter variations (“m-p-v”) was developed and implemented. It has been applied to the evaluation of uncertainty of reflood-related input parameters and models of CATHARE2 code. The research has been carried out in the framework of several international projects and co-operations, and has thus profited of the availability of large experimental databases and numerical resources, as well as of the valuable interactions with the involved International scientific community.